Primary competition visual

Intron AfriSpeech-200 Automatic Speech Recognition Challenge

$5 000 USD
Challenge completed over 2 years ago
Automatic Speech Recognition
430 joined
41 active
Starti
Feb 17, 23
Closei
May 28, 23
Reveali
May 28, 23
User avatar
GideonG
Zindi Ambassador to Nigeria
training_args = Seq2SeqTrainingArguments(...)
Notebooks · 8 May 2023, 17:45 · 2

Please someone should help me I can't run this cell in the starter notebook.

from transformers import Seq2SeqTrainingArguments

training_args = Seq2SeqTrainingArguments(

output_dir="./dsn_afrispeech", # change to a repo name of your choice dsn_afrispeech

per_device_train_batch_size=8,

gradient_accumulation_steps=1, # increase by 2x for every 2x decrease in batch size

learning_rate=1e-5,

warmup_steps=500,

max_steps=4000,

gradient_checkpointing=True,

fp16=True,

evaluation_strategy="steps",

per_device_eval_batch_size=8, #try 4 and see if it crashes

predict_with_generate=True,

generation_max_length=225,

save_steps=500,

eval_steps=500,

logging_steps=25,

report_to=["tensorboard"],

load_best_model_at_end=True,

metric_for_best_model="wer",

greater_is_better=False,

#push_to_hub=True,

)

This is the error I get below, I tried to solve it to no avail

NameError: name 'PartialState' is not defined

Discussion 2 answers

Can you paste the full stack trace of the error?

8 May 2023, 21:39
Upvotes 0
User avatar
Siwar_NASRI

Solved on the discussion 'Incompatibility': downgrade the huggingface transformers

13 May 2023, 21:19
Upvotes 0